MCP Reasoner
Implements beam search and thought evaluation for structured problem-solving, enabling exploration of multiple solution paths in c
This MCP reasoner, developed by Jacck, provides a systematic approach to problem-solving using beam search and thought evaluation. Built with TypeScript and leveraging the Model Context Protocol SDK, it offers capabilities for exploring multiple solution paths, scoring thoughts, and managing tree-based reasoning states. The reasoner implements a configurable beam width, depth-limited search, and statistical analysis of the reasoning process. It's designed for tackling complex problems that require step-by-step analysis, such as mathematical puzzles, logical deductions, and multi-stage planning tasks. This implementation is particularly valuable for enhancing AI assistants' problem-solving abilities in scenarios requiring structured thinking and exploration of multiple solution strategies.
Source
Repository: https://github.com/jacck/mcp-reasoner
Maintain MCP Reasoner?
Let people know it's listed here — add the badge (live metrics, light/dark aware) or a plain link to your README or docs.
[](https://getagentictools.com/mcp/jacck-mcp-reasoner?ref=badge) Related mcp
DeepContext (Semantic Code Search) Symbol-aware semantic search for large codebases
Ultra (Multi-AI Provider) Unified server providing access to OpenAI O3, Google Gemini 2.5 Pro, and Azure OpenAI models with automatic usage tracking, cost e
Wikipedia Provides a bridge to Wikipedia's knowledge base, enabling search, retrieval, and analysis of articles through the Wikipedia API fo
Qdrant Docs Rag Real-time documentation context through vector-based search and retrieval via Qdrant.
MongoDB Query and analyze MongoDB databases.
HeyClaude Read-only MCP server providing access to a curated registry of 1,300+ Claude agents, MCP servers, skills, tools, prompts, and guid